AlnBlock * Align_strings_ProteinSmithWaterman(char * one,char * two,CompMat * comp,int gap,int ext,DPRunImpl * dpri)
{
Sequence * one_s;
Sequence * two_s;
AlnBlock * out;
/* error check the strings? */
one_s = new_Sequence_from_strings(NULL,one);
if( one_s == NULL ) {
warn("Cannot make new sequence...\n");
return NULL;
}
two_s = new_Sequence_from_strings(NULL,two);
if( two_s == NULL ) {
warn("Cannot make new sequence...\n");
return NULL;
}
out = Align_Sequences_ProteinSmithWaterman(one_s,two_s,comp,gap,ext,dpri);
free_Sequence(one_s);
free_Sequence(two_s);
return out;
}
int main(int argc,char ** argv)
{
AlignGeneModelParam * agmp;
SeqAlign * sal;
Sequence * seq;
CompProb * cp;
CodonTable * ct;
DnaProbMatrix * dm;
dm = DnaProbMatrix_from_match(0.8,NMaskType_VARIABLE);
ct = read_CodonTable_file("codon.table");
cp = read_Blast_file_CompProb("wag85");
sal = SeqAlign_alloc_std();
seq = new_Sequence_from_strings("new1","ATGGGG");
add_SeqAlign(sal,seq);
seq = new_Sequence_from_strings("new2","ATGGGT");
add_SeqAlign(sal,seq);
agmp = std_AlignGeneModelParam(cp,dm,ct);
printf("Codon 1 %f vs %f Codon 2 %f v %f \n",
coding_probability_AlignGeneModel(sal,2,agmp),
non_coding_probability_AlignGeneModel(sal,2,agmp),
coding_probability_AlignGeneModel(sal,5,agmp),
non_coding_probability_AlignGeneModel(sal,5,agmp)
);
}
Sequence * swapped_Sequence(Sequence * reference,HitList * hl)
{
Sequence * out;
int i;
AlnColumn * alc;
char buffer[512];
char * swapped;
int coord;
int swap;
sprintf(buffer,"swapped_%s",reference->name);
out = new_Sequence_from_strings(buffer,reference->seq);
for(i=0;i<reference->len;i++) {
out->seq[i] = tolower(reference->seq[i]);
}
/* calloc sets things to 0 be defn */
swapped = calloc(reference->len,sizeof(char));
for(i=0;i<hl->len;i++) {
fprintf(stderr,"Using... %s\n",hl->pair[i]->target->name);
for(alc =hl->pair[i]->aln[0]->alb->start ;alc != NULL;alc = alc->next) {
/* fprintf(stderr," Looking at %d,%d\n",alc->alu[0]->end,alc->alu[1]->end);*/
if( strcmp(alc->alu[0]->text_label,"SEQUENCE") == 0 &&
strcmp(alc->alu[1]->text_label,"SEQUENCE") == 0 ) {
/* fprintf(stderr,"MATCH Looking at %d,%d\n",alc->alu[0]->end,alc->alu[1]->end); */
coord = alc->alu[0]->end;
swap = hl->pair[i]->target->seq[alc->alu[1]->end];
if( swapped[coord] == 1 ) {
warn("Reswapping position %d to %c from %c",coord,swap,out->seq[coord]);
} else if( toupper(swap) != toupper(out->seq[coord]) ) {
info("Swapping position %d from %c to %c",coord,out->seq[coord],swap);
}
out->seq[coord] = toupper(swap);
swapped[coord] = 1;
}
}
}
ckfree(swapped);
return out;
}
static Wise2HSP_HSP_ts_iterator
impl_Wise2HSP_HSPmanagerFactory_scan_query_iterator(impl_POA_Wise2HSP_HSPmanagerFactory * servant,
Wise2HSP_Sequence * query,CORBA_short max_size,
CORBA_Environment * ev)
{
Wise2HSP_HSP_ts_iterator retval;
Wise2HSP_HSPmanager* corba_hspm;
Wise2_Sequence * wise2_seq;
LinearHSPmanager * lm;
HSPScanInterfacePara para;
int size;
int i;
if( servant->is_verbose == 1 ) {
fprintf(stdout,"going to do %s\n",query->name);
}
wise2_seq = new_Sequence_from_strings(query->name,query->seq);
para.max_results = max_size;
lm = (*servant->hsi->scan_query)(servant->hsi->data,wise2_seq,¶);
/*
fprintf(stderr,"Sanity check in corba\n");
for(i=0;i<lm->len;i++) {
fprintf(stderr,"Got %s %d\n",lm->set[i]->hsp[0]->target->name,i);
}
*/
if( servant->is_verbose == 1 ) {
fprintf(stdout," ... retrieved HSP manager size %d\n",lm->len);
}
if( servant->is_verbose == 1 ) {
fprintf(stdout," ... remapped HSP manager\n");
}
if( servant->is_verbose == 1 ) {
fprintf(stdout," ... freed data, about to return iterator\n");
}
retval = impl_Wise2HSP_HSP_ts_iterator__create(servant->poa,lm,max_size,ev);
return retval;
}
static Wise2HSP_HSPmanager*
impl_Wise2HSP_HSPmanagerFactory_scan_query(impl_POA_Wise2HSP_HSPmanagerFactory *servant,
Wise2HSP_Sequence* query,
CORBA_Environment *ev)
{
Wise2HSP_HSPmanager* retval;
Wise2_Sequence * wise2_seq;
LinearHSPmanager * lm;
HSPScanInterfacePara para;
int size;
if( servant->is_verbose == 1 ) {
fprintf(stdout,"going to do %s\n",query->name);
}
wise2_seq = new_Sequence_from_strings(query->name,query->seq);
para.max_results = 250;
lm = (*servant->hsi->scan_query)(servant->hsi->data,wise2_seq,¶);
retval = corba_manager_from_wise_manager(lm);
if( servant->is_verbose == 1 ) {
fprintf(stdout," ... remapped HSP manager\n");
}
Wise2_free_LinearHSPmanager(lm);
if( servant->is_verbose == 1 ) {
fprintf(stdout," ... freed data, about to return\n");
}
return retval;
}
ErrorSequence * make_ErrorSequence(Sequence * seq,Probability subs,Probability insertion,Probability deletion)
{
ErrorSequence * out;
char * seq_buffer;
int buf_len;
int i,j;
double rnd;
char c[2];
SequenceError * se;
i=j=0;
out = ErrorSequence_alloc();
out->ses = SequenceErrorSet_alloc_std();
c[1] = '\0';
buf_len = seq->len *3;
seq_buffer = (char*)calloc(buf_len,sizeof(char));
for(i=0;i<seq->len;i++) {
if( j >= buf_len ) {
warn("run out of temporary buffer in error sequence");
break;
}
rnd = random_0_to_1();
if( rnd > deletion ) {
/* yes this base is here */
rnd = random_0_to_1();
if( rnd < subs ) {
rnd = random_0_to_1();
if( rnd < 0.25 ) {
c[0] = 'A';
} else if ( rnd < 0.5 ) {
c[0] = 'T';
} else if ( rnd < 0.75 ) {
c[0] = 'G';
} else {
c[0] = 'C';
}
se = SequenceError_alloc();
se->type = SeqErrorSubstitution;
se->start = j+1;
se->end = j+1;
se->replaced_bases = stringalloc(c);
add_SequenceErrorSet(out->ses,se);
seq_buffer[j++] = c[0];
} else {
seq_buffer[j++] = seq->seq[i];
}
} else {
/* there has been a deletion */
se = SequenceError_alloc();
se->type = SeqErrorDeletion;
se->start = j+1;
se->end = j+1;
c[0] = seq_buffer[i];
se->suspected_deletion = stringalloc(c);
add_SequenceErrorSet(out->ses,se);
}
rnd = random_0_to_1();
if( rnd < insertion ) {
rnd = random_0_to_1();
if( rnd < 0.25 ) {
c[0] = 'A';
} else if ( rnd < 0.5 ) {
c[0] = 'T';
} else if ( rnd < 0.75 ) {
c[0] = 'G';
} else {
c[0] = 'C';
}
se = SequenceError_alloc();
se->type = SeqErrorInsertion;
se->start = j+1;
se->end = j+1;
se->inserted_bases = stringalloc(c);
add_SequenceErrorSet(out->ses,se);
seq_buffer[j++] = c[0];
}
}
seq_buffer[j] = '\0';
out->seq = new_Sequence_from_strings(seq->name,seq_buffer);
return out;
}
